Review a false ban
The procedure for working one ban: read the evidence, decide, lift it, then fix the cause so the next player does not hit it.
Last updated
A player says they were banned for nothing. This page is the order to do things in, the two lists that tell you how much weight to put on the detection name, and what happens when you lift a ban.
The procedure
Find the ban and open it
The player has the ban id off their rejection screen, and the Bans List searches on it. Opening the ban gives you the reason, the identifiers, the evidence and the Additional info panel.
Read Additional info
Config Option names the detection and the rest of the panel names the value that fired it. Reading a detection covers every field. A ban with no panel at all was filed by hand, so there is no detection to review and the question is which of your admins filed it.
Weigh the detection name
Check it against the two tables below. If it is on the first list, the detection fired on something the server can prove and you can usually act straight away. If it is on the second, the detection name tells you very little on its own and the evidence decides.
Open the evidence
A screenshot or recording is a Download link on the ban and on the Discord card. For anything on the second list this is the step that settles it. A capture that came back as a frame the game never drew says so rather than showing you a black image, so a blank capture is not evidence of anything either way.
Decide, then lift the ban
Unbanning needs the bans remove permission. If the ban carries a detection panel the list also offers a second button that resolves the detection as a false positive and then unbans, which is the one to use when you intend to stop it recurring as well.
Fix the cause
Lifting the ban helps one player. The panel almost always names something you can whitelist, and whitelisting it is what stops the next twenty.
Detections you can usually act on
These fire on something the server can prove: a value the game cannot produce, an exact match against a list you control, or a request a normal client never sends. A hit here is almost always the player. Check the evidence anyway when the ban is permanent.
| Detection | Why it is hard to fire by accident |
|---|---|
AntiAimbot | Catches aimbot software, which ordinary play does not produce. The softer aim assist check, AntiAimAssist, ships off. |
AntiSilentAim | A hit registered while the weapon was never pointed at the target. |
AntiMagicBullet | A bullet that hits through a solid wall. |
AntiGiveWeapons | A weapon appearing in an inventory the server never put it in. |
AntiRemoveWeapons | A weapon leaving an inventory the server did not take it from. |
AntiStartResources | A client asking the server to start a resource. |
AntiStopResources | A client asking the server to stop a resource. |
AntiSpoofBulletShot | Faked shot data, which ordinary play does not produce. |
AntiBlacklistWeapons | An exact match against a list you control. |
EntitiesModelProtection | An exact model match against a list you control, in the default blacklist mode. In allowlist mode this does not hold: any script-spawned vehicle you did not list fires it, so read the evidence first. |
AntiSpawnIsolatedVehicle | A vehicle created outside any normal spawn path. |
AntiDamageExploits | Damage values the game cannot produce. |
AntiSpoofedDamage | Damage attributed to a weapon that did not fire it. |
Detections to read the evidence on first
These are not unreliable, but their trigger can also be produced by your own scripts, a weak machine or a bad connection. Read the evidence and the panel before you punish, and when one of them fires repeatedly on ordinary players, treat that as a config problem rather than a cheating problem.
| Detection | What the innocent explanation looks like |
|---|---|
ClipboardScanner | Ordinary words can match its word list. The list ships with print, script and getname on it, so sprint and a sentence about a script both match. |
AntiScreenOcr | Catches cheat text shown on screen. Text your own UI draws can match its word list, though ordinary text rarely does. Ships off. |
AntiExecutor | Overlays and capture tools can look like an injector. The prose reasons are the noisiest output in the product, so read the evidence before acting on one. |
AntiNuiTamper | Its Timeout type can fire on a weak machine with no cheat running. |
AntiGodmode | Health scripts that heal instantly can look the same as godmode. |
AntiSpawnWeapons | Any custom weapon you have not registered reads as an unknown weapon appearing. |
AntiSpeedHack | Sprint modifiers, parkour scripts and vehicle ejection all produce real speed. Ships off. |
AntiNoclip | Falling through unloaded map geometry looks the same as walking through it. |
AntiHijackVehicle | Garage scripts and vehicle resets move a player instantly on purpose. |
AntiParticleAi | Scripted effects on a busy scene can cross the spam rules without anyone cheating. |
AntiExplosionAi | Scripted explosions in a job or an event can cross the spam rules the same way. |
AiServerEventsProtection | The highest false-positive surface on any framework server. Read the event name first. |
EntitiesSpamProtection | An inventory dropping a large stash, or a garage pulling several cars, counts as spam. |
AntiResourcesInjection | Catches code injected into a running resource, so a loader or a library can trip it. The log stamps a Resource field. Read it before acting. |
What lifting a ban does
An unban does more than flip a row on your list. Three things follow from it.
| What changes | Detail |
|---|---|
| The player can connect | The connection check only matches bans that have not been reversed, so the next join goes through without anything being restarted. |
| An automatic network ban comes down with it | If the detection was one that escalates, the network-wide ban filed from this ban is withdrawn in the same action. It stays only when another server still holds the same person on another hard-signal ban. A network ban a staffer filed by hand is never touched by your unban. |
| The trust score is restored | The deduction the ban applied is reversed, and the globally banned flag is cleared unless an active network ban still covers the player. |
What does not change is the config. The detection is still set exactly as it was, so the next player who does the same thing is banned the same way.
Then fix the cause
The panel almost always names something you can whitelist: the resource that created an entity, the event that was blocked, the model, or the weapon. Whitelisting that one thing is better than raising a threshold, because a threshold change weakens the detection for everybody while a whitelist entry weakens it for exactly the thing you decided to trust.
Every toggle, action, threshold and whitelist is on the detection's own card in Protection Rules. If a detection has sub-checks and only one of them is wrong for your server, turn off that single Types entry rather than the whole detection. And if you are about to switch something straight to ban that you have never run here, set it to kick first and watch the activity log for a few days instead.